Southern Baltic sea level extremes: tide gauge data, historic storms and confidence intervals

Knowledge of extreme sea levels is important when planning housing developments and other infrastructure in coastal locations. The natural science basis of such plans are often return level-return period plots derived from tide gauge records that typically stretch from a few decades to a century. Coastal planners, however, often require return levels associated with return periods that are much longer than these tide gauge records in their planning. Moreover, return level estimates are known to be sensitive to outliers and can have significant biases. Here, we quantify different confidence intervals that are applicable to such return level estimates, and discuss their usability for coastal planning in the context of historic data from the Baltic Sea flood in 1872. Two types of commonly used confidence intervals are found to be too narrow to capture a plausible range that includes the 1872 Baltic Sea flood. A parametric bootstrapping method is then introduced, which gives a reasonable range even when this extreme flood is considered.